All cases of CRS are not created equal

With that specific CAR, Dr Barrett said they observed a MAS pattern—IFNγ, IL-10, IL-6, and IL-8, which are most elevated in grades 4 and 5 CRS.

“[S]o this pattern, and this clinical syndrome [CRS] was what we believe was driving toxicity in this model,” he said.

To figure out why this was happening, the investigators created 4-1BB CAR-mediated CRS in a mouse model.

The team took leukemia cells from the first patient treated and clinical T cells from her CAR product and put them in an NSG mouse model that they had used for preclinical development.

The investigators then measured cytokine production in the serum of animals 3 and 7 days post-treatment with CTL019.

“And nothing happened,” Dr Barrett said. “The mice didn’t get sick, they cleared their leukemia, and when you looked for cytokines, you found IFNγ, IL-2, and GM-CSF, but you did not find IL-6.”

The team had also included etanercept and tocilizumab in this model, but since the mice didn’t make the toxic cytokines, the antibodies didn’t do anything.

“So why did she get so sick but yet her cancer and her CAR T cells did not make these mice sick and not generate these cytokines?” Dr Barrett asked.

The investigators hypothesized that APCs—not the CAR T cells—were responsible for the toxic cytokines secreted.

“[I]t would be the CAR T-cell-mediated killing of leukemia which would induce this cytokine release from the antigen-presenting cell lineages,” Dr Barrett explained.

To test this theory, the investigators co-cultured CTL019 and Nalm-6 leukemia, with or without cells derived from peripheral blood monocytes.

The team found that IL-6 levels were elevated several logs when CAR T cells killed leukemia in the presence of the APCs.

On the other hand, co-culture of only CTL019 and Nalm-6 produced high levels of GM-CSF, IFNγ, IL-2, and IL-10 but no detectable IL-6 or IL-8.

Transwell in vitro experiments separating CTL019 and Nalm-6 from the APCs showed the same pattern.

The investigators thus confirmed that IL-6 is made by APCs in response to CAR-mediated killing of leukemia.

Nanostring profiling

The team then performed nanostring RNA analysis of separated cell populations recovered from that experiment.

They found that IL-6 and IL-8 are produced by APCs but not by CTL019. IL-2 and IFNγ are produced by CTL019 and not by APCs, and GM-CSF was produced from CTL019.

“There was a clear separation in cytokine production in this model,” Dr Barrett said.

The investigators also observed that the CTL019 nanostring profile was unaffected by proximity to the APCs and all the IL-6 they make.

“CART19 T cells did not seem to care, on a transcriptional level, that all this IL-6 was floating around,” Dr Barrett said.

In contrast, the APCs do change, he said, when CAR T cells are killing leukemia nearby.

“There are dozens and dozens of changes,” he said, “including many in chemokines and IL-6 and IL-8.”

The investigators performed multiple in vitro killing assays and found no difference in CAR T-cell killing potential in the presence or absence of the MAS cytokines.

They also performed peripheral blood analysis of patients experiencing CRS of grades 2 to 5. The team observed that clinical CRS may be divided into MAS and not-MAS patterns. In addition, they detected no IL-6 transcript in any of the CAR T cells isolated from these patients.

“I think we’re going to discover that cytokine release syndrome is a clinical entity that has multiple mechanisms,” Dr Barrett said. “And so it’s very important, when we are talking about our models and talking about our results, to be sure that we’re all speaking the same language.”